Statistics

More ESTs support the specific splicing site, this splicing information is more reliable. We can also use the lowest frequency of ESTs aligned to a specific splicing site as a parameter. The lowest frequency user specified is higher, the alternative splicing could be found is less. In the following tables, we list the number of alternative splicing genes in human and mouse with different lowest EST frequencies. (Table 1, Table 2)If only the splicing sites with at least two ESTs supported are considered, the amounts of alternative splicing sites for different organisms are stated as follows. (Table 3)

HS EST support	Exon Skpping	3’ AS	5’ AS	ME	Total
≧1	17418	12933	12879	270	43500
≧2	5800	3227	3213	133	12373
≧3	3093	1626	1520	87	6326
≧4	1996	947	882	64	3889
≧5	1429	653	582	52	2716

MM EST support	Exon Skpping	3’ AS	5’ AS	ME	Total
≧1	8068	6567	6473	119	21227
≧2	2772	1504	1488	65	5829
≧3	1433	676	702	42	2853
≧4	927	420	423	26	1796
≧5	638	305	287	18	1248

Organism	Exon Skipping	3’ AS	5’ AS	ME	Total
HS	5800	3227	3213	133	12373
MM	2772	1504	1488	65	5829
RN	158	145	162	0	465
DM	8	100	106	0	214
CE	7	50	63	1	121
AT	2	59	76	0	137

There is substantial variation in the certainty associated with these observations, due to the differing number of EST frequency associated with each of these AS sites. We list the number of AS sites and AS genes in human with different EST lowest frequencies. (Table 4)

Since the coverage of ESTs in genome is really rugged, a small number of genes being represented by large numbers of ESTs. This set might be biased towards medically relevant genes. When we compute the ratio of AS genes, the number of ESTs mapped into the same gene region should be considered. In the following table, we show that the genes associated with more ESTs are with higher alternative splicing possibility. (Table 5)

EST supprot	No. of EST	AS events	AS genes
≧1	858838	115915	26814
≧2	760737	33261	18490
≧3	693091	17886	13817
≧4	642163	11435	11426
≧5	599929	8780	9463

Table 5. Relation of Alternative splicing Gene associated with EST frequencies

Human ESTs coverage /Genes	Num of Genes	Alternative splicing genes	Ratio
≧ 2,000	23	23	100%
≧ 1,000	87	86	99%
≧ 800	124	121	98%
≧ 300	600	577	96%
≧ 100	3525	2880	82%
≧50	7193	5482	76%
≧10	19303	8956	46%
≧5	23878	9117	38%
≧2	29520	9119	30%
=0	26377	0	0%

As EST data collection continues, it even seems probably that alternative splicing variant may be observed for all genes . Yet, in Avatar, some genes are associated with many ESTs and no alternative splicing event been observed. For example, NM_000184.1 NM_000559.1, NM_015379.1, NM_002415.1, NM_018955.2, NM_009199.1 are all associated with more than 500 ESTs and no obvious alternative splicing event is detected. (Table 6)

Table 6. The genes involve more than 500 ESTs but no obvious alternative splicing

Organism	EST support	mRNA accession	Gene name
HS	1015	NM_000184.1	HBG2
HS	987	NM_000559.1	HBG1
HS	806	NM_015379.1	BRI3
HS	711	NM_002415.1	MIF
HS	678	NM_018955.2	UBB
MM	610	NM_009199.1	Slc1a1

Using bioinformatic approach to detect AS sites, we first need to map ESTs to different clusters. Traditionally, researchers use UniGene clustering information to classify ESTs. It saves computation time. Yet, it is not informative. Our approach aligns ESTs to genome and cluster ESTs according to their corresponding location at genome. Many ESTs in different UniGene clusters are aligned to the same genome area. For example, we find that UniGene cluster HS.433680, HS.432883, HS.305916 and HS.396617 are all mapped into gene TG(Thyroglobulin).